On-chip spectrometer employing pixel-count-modulated spectral channels and method of manufacturing the same

Abstract

An array of sensor pixels is formed on a substrate, and a signal processing unit is connected to the array of sensor pixels. The signal processing unit includes multiple spectral channels that are defined by a respective transmission curve of each optical filter of at least one associated sensor pixel. Each of the sensor pixels includes a stack of a respective photodetector and a respective optical filter. Each spectral channel receives an output signal from one or more sensor pixels including an optical filter having the same transmission curve. At least one spectral channel has a greater number of sensor pixels than another spectral channel among the multiple spectral channels. The different number of pixels for the spectral channels can be employed to compensate for variations of sensor efficiency as a function of wavelength. Adjustment to sensor gain can be minimized through use of different number of pixels for different spectral channels.

What is claimed is: 
     
       1. A spectrum sensor comprising:
 an array of sensor pixels located on a substrate; and 
 a signal processing unit including L spectral channels, L being an integer greater than 7, wherein: 
 each of the sensor pixels comprises a stack of a respective photodetector and a respective optical filter configured to pass light within a respective transmission curve; 
 for each integer k from 1 to L, a k-th spectral channel receives an output signal from the sensor pixels including an optical filter proving a k-th transmission curve; and 
 at least one spectral channel among the L spectral channels has a greater number of sensor pixels than another spectral channel among the L spectral channels; 
 wherein the spectrum sensor comprises at least one feature selected from: 
 (i) a first feature wherein a ratio of a maximum number of sensor pixels per spectral channel to a minimum number of at least one sensor pixel per spectral channel is in a range from 1.1 to 30; or 
 (ii) a second feature wherein at least four of the L spectral channels are used as reference pixels which have no filters, covered by only transparent layer and receives full spectrum of input light, wherein reference pixels are used for optimal integration time calculation; or 
 (iii) a third feature wherein each photodetector in the array of sensor pixels comprises a same semiconductor material having a same photosensitive junction structure and has a same device area; or 
 (iv) a fourth feature wherein each optical filter is a plasmonic filter including a respective conductive material sheet and openings, particles, wires or pillars therethrough; or 
 (v) a fifth feature wherein for each integer k from 1 to L, each sensor pixel for the k-th spectral channel has a k-th detection efficiency, for a positive integer kp that is less than L+1, a kp-th detection efficiency is a maximum detection efficiency among all detection efficiencies of the sensor pixels, and at least one spectral channel among the L spectral channels has a greater number of sensor pixels than a total number of sensor pixels for the kp-th spectral channel. 
 
     
     
       2. The spectrum sensor of  claim 1 , wherein:
 the array of sensor pixels is arranged as a rectangular M×N array of sensor pixels; 
 M is an integer greater than 4; and 
 N is an integer greater than 4. 
 
     
     
       3. The spectrum sensor of  claim 1 , wherein the at least one feature comprises the first feature. 
     
     
       4. The spectrum sensor of  claim 1 , wherein: more than 10 pixels are used for black pixels which have no filters, covered by metal and receive no light; and black pixels are used for dark value compensation. 
     
     
       5. The spectrum sensor of  claim 1 , wherein the at least one feature comprises the second feature. 
     
     
       6. The spectrum sensor of  claim 1 , wherein at least one of the L spectral channels receives an output signal from only 1 or 2 sensor pixels. 
     
     
       7. The spectrum sensor of  claim 1 , wherein the at least one feature comprises the third feature. 
     
     
       8. The spectrum sensor of  claim 1 , wherein the at least one feature comprises the fourth feature. 
     
     
       9. The spectrum sensor of  claim 1 , wherein the at least one feature comprises the fifth feature. 
     
     
       10. The spectrum sensor of  claim 9 , wherein, for each integer from 1 to L, a product of the k-th detection efficiency and a total number of at least one sensor pixel for the k-th spectral channel is within a range from 50% to 200% of a product of the kp-th detection efficiency and a total number of at least one photodetector spectral channel for the kp-th spectral channel. 
     
     
       11. The spectrum sensor of  claim 10 , wherein the signal processing unit comprises an array of signal attenuators configured to attenuate a signal for a respective one of the L spectral channels. 
     
     
       12. The spectrum sensor of  claim 11 , wherein:
 the signal processing unit comprises a calibration table including attenuation factors for each of the L spectral channels; and 
 each of the attenuation factors is within a range from 0.5 to 1.0. 
 
     
     
       13. A method of fabricating a spectrum sensor, comprising:
 forming an array of sensor pixels on a substrate; and 
 electrically connecting a signal processing unit to the array of sensor pixels, wherein: 
 the signal processing unit includes L spectral channels, L being an integer greater than 7; 
 each of the sensor pixels comprises a stack of a respective photodetector and a respective optical filter configured to pass light within a respective transmission curve; 
 for each integer k from 1 to L, a k-th spectral channel receives an output signal from the sensor pixels including an optical filter providing a k-th transmission curve; 
 at least one spectral channel among the L spectral channels has a greater number of sensor pixels than another spectral channel among the L spectral channels; and 
 wherein a ratio of a maximum number of sensor pixels per spectral channel to a minimum number of at least one photodetector spectral channel per spectral channel is in a range from 1.1 to 30. 
 
     
     
       14. The method of  claim 13 , wherein:
 the minimum number of at least one sensor pixel per spectral channel is for a spectral channel connected to an optical filter having a peak transmission wavelength within a wavelength range from 500 nm to 600 nm; and 
 the maximum number of sensor pixels per spectral channel is for a spectral channel connected to an optical filter having a peak transmission wavelength within a wavelength range from 750 nm to 1,200 nm. 
 
     
     
       15. A method of fabricating a spectrum sensor, comprising:
 forming an array of sensor pixels on a substrate; and 
 electrically connecting a signal processing unit to the array of sensor pixels, wherein: 
 the signal processing unit includes L spectral channels, L being an integer greater than 7; 
 each of the sensor pixels comprises a stack of a respective photodetector and a respective optical filter configured to pass light within a respective transmission curve; 
 for each integer k from 1 to L, a k-th spectral channel receives an output signal from the sensor pixels including an optical filter providing a k-th transmission curve; 
 at least one spectral channel among the L spectral channels has a greater number of sensor pixels than another spectral channel among the L spectral channels; 
 at least one of the L spectral channels receives an output signal from 10 or more sensor pixels; 
 at least another of the L spectral channels receives an output signal from only 1 or 2 sensor pixels: and 
 each photodetector in the array of sensor pixels comprises a same semiconductor material having a same photosensitive junction structure and has a same device area. 
 
     
     
       16. The method of  claim 15 , wherein:
 each optical filter is a plasmonic filter including a respective metal sheet and openings therethrough; 
 for each integer k from 1 to L, each sensor pixel for the k-th spectral channel has a k-th detection efficiency; 
 for a positive integer kp that is less than L+1, a kp-th detection efficiency is a maximum detection efficiency among all detection efficiencies of the sensor pixels; and 
 at least one spectral channel among the L spectral channels has a greater number of sensor pixels than a total number of sensor pixels for the kp-th spectral channel. 
 
     
     
       17. The method of  claim 16 , wherein, for each integer from 1 to L, a product of the k-th detection efficiency and a total number of at least one sensor pixel for the k-th spectral channel is within a range from 50% to 200% of a product of the kp-th detection efficiency and a total number of at least one photodetector spectral channel for the kp-th spectral channel. 
     
     
       18. The method of  claim 17 , wherein:
 the signal processing unit comprises an array of signal attenuators configured to attenuate a signal for a respective one of the L spectral channels; 
 the signal processing unit comprises a calibration table including attenuation factors for each of the L spectral channels; and 
 each of the attenuation factors is within a range from 0.5 to 1.0.